Agentic Shopping Makes Conscious Consumerism Actually Work

When agents do the shopping, transparency becomes the best marketing

There’s a certain amount of fatigue around conscious consumerism these days. Do I really have to learn about fabric manufacturing and labor practices in Indonesia to buy a shirt? Do I really need to learn about water management practices in California to buy a bag of almonds?

There are lots of ways in which people have preferences not only about the products they buy, but also about how the products were made. Animal welfare is one example, but people also care about things like labor practices, sustainability, product safety over and above regulation (e.g. microplastics), or even what political tribe the company aligns itself with.

Over the last few decades, modern supply chains have adjusted somewhat to meet this demand for better-made products, and modern consumers have clearly shown a significant willingness to pay for products that align with their values. But conscious consumerism fatigue happens because there’s too much to know about too many companies and industries. And many consumers feel burned after a company they thought they could trust betrays that trust in some way. What people want is for the companies they buy from to just do things better by default.

Companies face the mirror image of the consumer’s problem. I’m confident that there are many people in many businesses trying to find ways to do things better within the economic limitations of their industries. But consumers have a limited ability to tell when a company really is doing better, and therefore a limited ability to reward it. A company can invest heavily in marketing to tell consumers what it’s doing, but marketing is a difficult skill in its own right, entirely separate from the work of actually doing things better. And there’s often little stopping a competitor from making a similar claim without actually doing the hard work.

So we end up in an equilibrium where good actors aren’t fully compensated for creating positive externalities, bad actors engage in “X-washing,” and consumers are left somewhat unsatisfied with their ability to consciously consume. Clearly, there’s room for improvement.

When AI shops

AI agents could change this equilibrium because they remove the attention bottleneck.

As our AI agents become more capable, and learn more about us, one natural endpoint is that a significant amount of consumer purchasing in the future is going to be done through agents. My AI is going to know what I like, and I’ll be able to tell it (or it will just know) that I need new socks, a nice TV for my living room, or a restock of eggs.

This could change the economy in all sorts of ways, but one of the more hopeful outcomes is that it could solve the current problems with conscious consumerism, and make it way easier for companies to be incentivized to do what consumers actually want.

In our current equilibrium, since consumers have so little ability and bandwidth to actually look at what’s going on, businesses don’t have an incentive to report. If almost no one reports, companies that remain opaque look completely normal. A company that discloses more takes on more risk by making its flaws visible, without much reason to think consumers will reward it.

As AI agents start doing more of our purchasing, a new equilibrium could emerge. It begins with companies that are already doing things better surfacing more data and knowing that agents will actually see and understand it. This will become an alternative to brand marketing, or the AI version of SEO. Then, as more companies report and it becomes clear that transparency gets rewarded, the act of non-disclosure itself becomes a negative signal. If it’s known that a company could disclose something positive and benefit from it, but it doesn’t, the natural assumption will be that the company thinks its customers won’t like the answer. Once agents start rewarding transparency, the gravitational pull toward opacity reverses.

Given that AI’s comparative advantage is sifting through mountains of unstructured data, there needn’t be any standardization on what companies report or how they report it. They can just report whatever they think consumers care about, whether that’s price, carbon footprint, or whether upstream suppliers have ever said anything mean about their favorite politician. AI purchases will then reward them in proportion to how well they know their customer, and how they’re doing by their customer’s values.

The hypothesis is that the market for these kinds of premium products will grow when there’s a supply-demand match between what people want, and what companies actually do. Even if AI doesn’t make consumers care more, it could make it much easier for consumers to act on what they already care about without having to pay attention themselves, which would expand the market for products made in specific ways.

From vibes to data

I’ve written before about who captures the welfare premium in the animal product market. Right now, it’s brands like Vital Farms (the second biggest egg brand in the country by revenue), Whole Foods, and ButcherBox that command significant premiums through selling better-made products. The reason these companies are able to capture the premium is because they have consumer trust.

Consumer trust is pretty vibes-based right now, which is out of necessity. The average Whole Foods customer probably couldn’t tell you many specifics about why Whole Foods products are better, they just know that Whole Foods will get them the good stuff.

The market shift I’m describing represents a transfer of trust from brands to AI agents, and the basis of this trust shifts from vibes to data. There are some entities right now that try to provide more data-based trust, like third-party certifiers. But they don’t command premiums, because vibes-based trust is more what the current low-bandwidth consumer demands.

It’s hard to envision what this kind of radical transparency would look like given where we are today, but here is one way to illustrate it. Your agent gets to know you over time. It learns that you care about how animals are treated, along with a bunch of other things, and understands how much you can afford to spend. It can then evaluate products based on how well they match your preferences and budget.

And we’re not just talking about the company or the brand. The information could reach down to the individual product, or even the individual animal. When you buy a chicken breast, your agent could have access to information about the specific chicken it came from: its breeder flock and hatchery, the farm where it was raised, where it spent time within the barn, major events during its life, what it ate, whether it got sick, possibly even biomarkers related to stress, and how it was slaughtered. All this information could be backed up by auditable video footage and other data streams.

This is way more information than the consumer actually wants to know. But the information isn’t for the consumer, it’s for the agent. The consumer gets what they actually wanted, which is to know that their purchases align with their values.

This is pretty different from how supply chains work right now. But much of the physical monitoring infrastructure needed to make it possible, including cameras, sensors, machine vision, and individual animal tracking, is the same infrastructure companies will need to become AI-native anyway. I suspect the hardware transition will begin as part of the move toward Precision Livestock Farming, aimed mostly at improving efficiency and animal health. It will accelerate if and when companies realize that the same data can also give them a competitive edge with consumers.

Some might think that conscious consumerism is a fleeting trend of rich coastal elites and that ultimately this vision fails because people don’t care. Maybe. But I think it’s notable that the richest communities tend to be the ones with the most robust specialty markets. If AI makes us all rich, serving these kinds of consumers will become more important. Plus, I think part of the “people don’t care” story is that caring about which eggs you buy means both spending more money and also becoming an agricultural expert. With AI you just need to spend more money.

The customer is always right?

Some of today’s companies, accustomed to supply chains being opaque, might feel nervous about this vision. Consumers can be crazy and irrational in all sorts of ways, and giving them more say over how supply chains operate gives them more opportunities to shoot themselves in the proverbial feet. But the economist would point out that this new equilibrium is value-creating: it gives more people more of what they want, and it rewards the producers that provide it.

The truly utopian vision is that AI-mediated purchasing can make consumers more rational. Maybe when someone worries that giving a livestock animal a safe, lifesaving drug will make their child autistic, this fear comes from a sense of being kept in the dark over the quality of food going into their child’s body. And maybe once people trust their AIs to actually pay attention, this fear will be put to rest. I don’t know if this will happen. But even if it doesn’t, the company that resents its customers ultimately won’t succeed.

Today, doing things better is only half the battle. The other half is getting millions of busy strangers to notice, and plenty of good companies lose on that half. Once every person has a super-genius in their pocket deciding what to buy, the noticing gets handled. Convincing people you’re a good company stops being a separate activity from being one. What consumers care about and what businesses get paid for become, at last, the same thing. Call it a form of AI alignment.